CORRELATION BETWEEN POLLUTANTS EMISSION AND INHABITANT… · 2 CORRELATION BETWEEN POLLUTANTS...
Transcript of CORRELATION BETWEEN POLLUTANTS EMISSION AND INHABITANT… · 2 CORRELATION BETWEEN POLLUTANTS...
1
CORRELATION BETWEEN POLLUTANTS EMISSION AND
INHABITANT’S MORBIDITY:
SÃO PAULO CITY STUDY CASE
Gheisa R. T. Esteves
Sonia R.C.S. Barbosa Campinas, Brazil
Paper presented to the PRIPODE workshop on
Urban Population, Development and Environment Dynamics in Developing
Countries
Jointly organized by CICRED, PERN and CIESIN
With support from the APHRC, Nairobi
11-13 June 2007
Nairobi, Kenya
2
CORRELATION BETWEEN POLLUTANTS EMISSION AND
INHABITANT’S MORBIDITY:
SÃO PAULO CITY STUDY CASE1
Gheisa R. T. Esteves2 Sonia R.C.S. Barbosa3
1. Introduction
The humanity lives a moment in its history where it is ascending the
environmental concern. Not so long ago, the human kind believed that energy sources
were inexhaustible and that it could be consumed without attempting for problems like
scarcity and the harming effects of pollutants on human health and the environment. In
the end of XX century, a number of episodes pointed out the actual development model
have a extremely predatory character.
Air pollution is a growing problem in big urban centers. Cities like Los Angeles,
Tokyo, Mexico City and São Paulo has a huge vehicle fleet, not just for having high
concentration of population, but also due to the actual world socioeconomic conjuncture
(Goldenberg, 1998). Its conjuncture tends to prioritize the individual transport when
compared with the collective transports. When the focus is the developing countries, it
is necessary to add that, the collective transport usually has really bad conditions and it
ends up encouraging people to use individual transport.
In the beginning of the capitalism, during the industrial revolution, most of the
air pollution came from industries. After the introduction of rigid control to industrial
emission, the main source of emission became the vehicle fleet and therefore it is
nowadays the most responsible for the deleterious effects that air pollution has human
health.
The first episode of increase in the respiratory diseases morbidity and mortality
in a region due to the unexpected increases of pollutants concentrations was notified in
1 This paper is part of the doctoral research that Gheisa Roberta Telles Esteves has been developing under the supervision of Sonia R. C. S. Barbosa with financial support of National Council for Scientific and Technological Development (CnPq), Brazil. 2 PhD Student in Energy Planning - Mechanical Engineering Faculty – State University of Campinas (UNICAMP). Email: [email protected]. 3 Associate Researcher of Environmental Studies Center (NEPAM) at State University of Campinas (UNICAMP), Brazil.
3
1930, in an area between the cities of Huy and Liége (Belgium). The two cities are
located in the Muse Valley, which is a region with high concentration of industries. In
that occasion, the numbers of respiratory morbidity and mortality went up due to a
junction of some factors: adverse meteorological conditions (among them the absence
of wind flows), that made difficult for the pollutants to be dispersed. Other similar
events occurred some years later, in Pennsylvania (1948) and the most notorious and
serious one, in London (1952). In the “London Fog” episode, there was, approximately,
4.000 deaths and some thousands visits to the emergency rooms in just 3 days. Most of
the hospitals visits and deaths were caused by thermal inversion that made difficult the
dispersion of the pollutants emitted by the industries.
Due to the events listed above, it was created the “Clean Air Acts” in Europe
and the establishment of air quality standards in the United States. In developing
countries, especially in São Paulo city and Metropolitan area, not so serious episodes
motivated the government to adopt control measures for air quality and pollutants
emission both for transport and industrial sectors.
In São Paulo City, the LPAE (Atmospheric Pollution Laboratory) was the first to
study the effects of air pollution on human health, animals and plants. According to
LPAE, in São Paulo city, the risk of death by respiratory and cardiovascular diseases
increases 12% in days with high levels of air pollution. The most susceptible part of the
population is children and elderly people, besides from people with chronic respiratory
diseases (Saldiva, Braga&Pereira, 2001).
It’s well known the actual pollutants levels are bellow from the numbers
registered some decades ago (CETESB, 2006). However, those levels still are harming
to the population living in the area. It’s out of question that there was a mitigation of
those effects, but they remain causing problems on people’s health. In fact, any
pollutant level will have some adverse effects on the population.
Based on the problems remarked in the previous paragraphs, the objective of the
study is to establish the relation between air pollutants concentration and the number of
children visiting to hospitals in São Paulo. In the first part of the study, it will be
established the spearman correlation between the number of people going to public
health clinics monthly as well the monthly variation of each pollutant. The age groups
will be divided as follow: less than 1 year old, between 1 and 4 years old, between 5 and
9 years old, between 10 and 14 years old, between 15 and 19 years old, between 20 and
59 years old, between 60 and 69 years old, between 70 and 79 years old and more than
4
80 years old. Then, it will be modeled the group of children under 1 year old and
between 1 and 4 years old to using a generalized additive model, supposing that it
follows a Poisson distribution. In the final part it will be presented conclusions taken
from the results obtained.
2. Air Pollution and Impacts on Human Health: The Case of São Paulo
City
The air quality in São Paulo city is determined not just by its topographic
characteristics and meteorological conditions but also by the way the vehicles circulate
in the city and the localization of industries (Cetesb, 2004). Also the urban development
process occurred in the city after the fifties is responsible for the air quality problems,
because it end up causing the creation of a heat island in the area (Carmo, 1995, Cetesb,
2005).
In 2004, São Paulo city had its vehicle fleet composition as it’s shown on Figure
1. Most of the fleet is composed by vehicles and motorbikes. Even though, buses
corresponds to just 3% and trucks to 8% of the fleet, it’s necessary to take into
consideration that its daily circulation is much higher than of the other vehicles.
Caminhonetas
23%
Onibus
3%
Motos
25%
Automoveis
41%
Caminhoes
8%
Figure 1 – Organization of São Paulo City Vehicle Fleet – 2004
Source: Prodesp/Detran (2006)
Concerning the fuels used by the fleet, 77% of the vehicles run with gasoline,
16% with alcohol and 6% with diesel. It is important to say that the vehicles which run
with gasoline and alcohol follow the Otto cycle and the main pollutants emitted by them
5
are carbon monoxide, nitrogen oxides, hydrocarbons and aldehydes. In the case of
vehicles that run with gasoline, it also emits particulate matter and sulphur oxides. In
the vehicles moved by diesel the main pollutants emitted are particulate matter, nitrogen
oxides and sulphur oxides (Alvarez, 2002).
Facing the fact of the great gravity of the air pollution problem, it was created by
the government in 1980 a program fixing limits of pollutants emission. PROCONVE –
Programa de Controle da Poluição do Ar por Veículos Automotores (Vehicles Air
Pollution Control Program) was created based on international experiences and
determine maximum limits to vehicle and engines emissions. With the implementation
of this program, in 2000, the pollutants emissions reduced in 90% when compared with
1986.
CETESB is the institution responsible for monitoring and controlling the air
quality in São Paulo. According to the institution, air quality is measured by the
quantification of the harmful substances presents in the air. The pollutants who integrate
the air quality standard are total particles in suspension, inhale particles, smoke, sulphur
dioxide, nitrogen dioxide, carbon monoxide and ozone. On Table 1 is listed the scales
used for the air quality standard, and on Table 2 the effects that each scale has on human
health.
Table 1 – Scale and Air Quality Standard
Quality Index
PM10 -
(µg /m³)
O3 -
Ozone (µg /m³)
CO - Carbon Monoxide
(ppm)
NO2 - Nitrogen
Dioxide (µg /m³)
SO2 -
Sulphur Dioxide (µg /m³)
GOOD 0- 50 0-50 0-80 0-4,5 0-100 0-80REGULAR 51 - 100 50-150 80-160 4,5-9 100-320 80-365INADEQUATE 101 - 199 150-250 160-200 9'-15 320-1130 365-800BAD 200 - 299 250-420 200-800 15-30 1130-2260 800-1600REALLY BAD 300 - 399 >420 >800 >30 >2260 >1600
Source: CETESB, 2006.
6
Table 2 – Effects of Each Scale on Human
Health
Quality Effects on Human Health
Good Almost no risk to human health.
Regular Children, elderly and people with respiratory and/or cardiovascular diseases can present dry cough and fatigue. The rest of the population is not afected.
Inadequate
The whole population can present symptoms like dry cough, fatigue, pain in the eyes, nose and throat. Children, elderly and people with respiratory and/or cardiovascular diseases can present more intense adverse effects than the rest of the population.
Bad
The whole population can present an intensification of symptoms like dry cough, fatigue, pain in the eyes, nose and throat and still present lack of air and gasping breath. More serious effects to health will be presented by children, elderly and people with respiratory and/or cardiovascular diseases.
Really Bad
The whole population can present serious respiratory and cardiovascular diseases. Increase of premature deaths of children, elderly and people with respiratory and/or cardiovascular diseases.
Source: CETESB, 2006.
CETESB has a number of stations to measure the pollutants in different
locations around the city. The stations cover the most critical points of the city. Figure 2
shows the monitoring stations around the city, and on Table 3 there is information about
the variables measured in each one of the stations. It is important to say that just the
particulate matter is measured in all the monitoring stations.
8
Table 2 – Location and Pollutants Monitored at the Stations in São Paulo City and Metropolitan Area
LOCATIONCAPITAL MP10 SO2 NO NO2 NOX CO CH4 HCNM O3 UR TEMP VV DV P RADCambuci - Av. D. Pedro I, 100 - - - - - - - - - - - - - -
Centro1 - A. São Luiz com R. da Consolação - - - - -
Cerqueira Cesar2 - Av. Dr. Arnaldo, 725 - - - - - - - - -Congonhas - Al. Dos Tupiniquins, 157 - - - - - - -
Ibirapuera3 - Parque do Ibirapuera, setor 25 - -
Santana4 - Av. Santos Dumont, 1019 - - - - - - - - - - -
Lapa1/5/6 - Av. Embaixador Macedo Soares, 7995 - - - - - - - -
Moóca2 - Rua Bresser, 2341 - - - - - - - - - - -Nossa Senhora do Ó - R. Capitão José Aranha do Amaral, 80 - - - - - - - - - -Parque D. Pedro - P. D. Pedro II, 319
Penha1 - Av. Amador Bueno da Veiga, 2932 - - - - - - - - - - - - - -
Pinheiros8 - Rua Frederico Hermann Jr, 245 - - - - - - - - -Santo Amaro - Av. Padre José Maria, 355 - - - - - - - - - - -São Miguel Paulista - Rua Diego Calado, 166 - - - - - - - - -
Diadema - Rua Beijamin Constant, 3 - - - - - - - - - - - - -Guarulhos - E. E. do Bairro de S. Roque - P. CECAP - - - - - - - - - - - -Mauá - Rua Vitorino Del'Antonia, 150 - - - - - - - - - -Santo André - Centro, Rua das Caneleiras, 101-C - - - - - - - - - - -Santo André - Capuava, Rua Manágua, 2 - - - - - - - - - - - -São Bernardo do Campo - Rua Cásper Libero, 340 - - - - - - - - - - - -São Caetano do Sul - Rua Aurélia s/n (EMI F. Pessoa) V. Paula
Osasco5/6/7 - Av. dos Autonomistas c/ R. S. Maurício - - - - - - - - - -Taboão da Serra - Praça Nicola Vivilechio, 99 - - - - - - - - - - - - - -TOTAL RMSP MONITORS 23 7 9 9 9 11 2 2 12 4 4 13 13 2 1
INFORMATION MONITORED
METROPOLITAN AREA
Source: CETESB, 2006.
9
3. Impacts on Human Health – Previous Studies and Historical
Information
There are a number of studies and ways to measure the impacts of air pollution
on human health. The first studies started in the XX century as a consequence of the
events listed at the introduction. A different number of statistical techniques were
applied to establish this relation. The first studies (in the fifties) started using just simple
descriptive analysis from the data to point out the effects of air pollution on morbidity
and mortality. This study was done after the episode of the London fog. In the Sixties,
more sophisticated techniques were used, as the scientists’ analyzed daily number of
deaths and air pollutants concentration with correlation analysis and simple linear
regression. In both cases, the results obtained showed significant relationship between
the two variables.
In the Seventies, there were some innovation on the statistical framework, as the
scientists started to use multiple linear regression in the studies, as well as the addition
of meteorological variables on the models, besides from the trend and seasonality. The
use of multiple linear regression persisted until the Eighties when scientists started to
look for non-linear regression methods whose could explain those relation.
But it was just in the second half of the nineties and in the year 2000 that big
evolutions were made in the field. First it has begun to be used in the analysis auto-
regressive Poisson models and distributions. The event of a hospital visit or death
follows a Poisson distribution because the probability of a person to be admitted in a
hospital (or die) in a certain day or month, due to respiratory diseases is quite low.
Then in the years 2000, linear and additive generalized models started to be
applied to measure the relationship. The additive generalized models allow more
flexibility to describe the standard of association between the variables through smooth
functions. It gives the possibility of adjust the control variables (trend, seasonality and
meteorological factors) through a non parametric process, meanwhile on the linear
generalized models parametric process are used.
Most of the studies that deals with the effects of air pollution on human health
had been carried out in the developed world but countries like Brazil, Mexico and Chile
already have a significant number of studies in this matter (Cifuentes, 2001). Between
them, it can be mentioned a relevant study done by Martins et al (2002) where it was
10
analyzed the variation of the daily levels of carbon monoxide, sulphur dioxide, nitrogen
dioxide, ozone, particulate matter and the daily number of hospital visits of people with
64 years old or more in São Paulo from 1996 until 1999. The analysis was done using
additive generalized models. It was constructed individual models for each pollutant.
Martins concluded that an increase to 15 3mµg at the concentration of sulphur dioxide
will add up 14,5% more hospital visits of elderly people with pneumonia or influenza.
In the case of the ozone an increase of 38,8 3mµg will cause 8,07% more hospital
visits.
Studies conducted by Gouveia (2000) analyzed the relation between daily
concentration of pollutants and hospital visits of children (below 5 years old) and
elderly (65 years or more) in Rio de Janeiro and São Paulo. It was concluded that an
increase of 10 3mµg in particulate matter concentration would cause an increase in
hospital visits by 1,9% in Rio de Janeiro and 3,5% in São Paulo. Those are just two
between a numbers of studies done for São Paulo city related to the theme. Studies done
by Azevedo (1999), Azuaga (2000), Pereira (1999), Miraglia (1997) and Saldiva also
have given interesting contribution to the issues.
4. Methods
The methodology used to evaluate the impacts of air pollution on health of
children of São Paulo city was divided into 3 phases. Initially, it was established the
existing relation between the pollutants concentration and the child hospital visits with
respiratory diseases. The next step was to develop a model to estimate the importance of
this correlation. As already said before in this article, hospitals visits usually follows a
Poisson distribution, so the present study will use additive generalized models (in a
semi-parametric model) to relate the pollution with child hospitals visits. The model can
be described as written bellow (Simas, 2003; Grant, 2001, Woolson, 1987).
[ ]( ) ( )tjp
jjt xfYELn ∑
=+=
1α (Equation. 1)
Where
Y : Number of hospital visits;
11
t : Month;
[ ]tYE : Expected number of hospital visits at month t;
α : trend;
( )tjj xf : Group of arbitrary and non specified functions of the independent variables.
( )µPoissonYt ~ that suposing
The independent variables were smoothed with the splines technique. This
technique consists of divide the variable that is supposed to be predicted into pre-
defined intervals and then adjust a polynomial (usually a cubic one) to each one of these
intervals, in a way that make them be grouped smoothly.
The association between air pollution and hospital visits can be influenced by
other factors, known as confusion factors. Adding those factors into the modeling has
the purpose of giving to each model component its exact contribution. So a model of
this nature must contain as independent variables, not just the pollutants concentration,
but also controls for seasonality (through month and year indexes), temperature,
humidity and the series trend.
It is important to emphasize the existence of multicolinearity between the
pollutants itself and with the meteorological variables. The correlation method used in
the article is the Spearman correlation, as the modeling method applied is a non linear
and semi-parametric one so it’s necessary to use also a non linear correlation
framework.
As the final models were obtained for the two groups studied (children
bellow 1 year old and children between 1 and 4 years old), the relative risk were
calculated for them. The relative risk is the percentage of hospitals visits occurred due
to the concentration of certain pollutant. Its mathematical expression is presented bellow
(Miraglia, 1997; Martins, 2002):
( )( ) 1001 ×−× tPe β (Equation 2)
Where
β : pollutant or index coefficient;
12
tP : pollutant or index average.
And the confidence interval where the relative risk is inserted is expressed with
the following equation
( )[ ]ββ ep 96,1%95
±= eIC (Equation.3)
Where
%95IC : 95% confidence interval;
β : Pollutant or index coefficient;
( )βep : Pollutant or index standard error.
Based on the results from the relative risk, it’s possible to determine the amount
of children visiting hospitals due to air pollution.
4.1 Air Pollution Data
Monthly pollutants concentration data was obtained with CETESB for the period
between January 1998 and March 2006. It was collected data for the following
pollutants: PM10, NO2, CO and SO2 for each monitoring station in São Paulo city.
Actually, there are 13 monitoring stations at São Paulo city. Only PM10 is
monitored in all 13 stations; SO2 is measured in 6 of them; 7 measure NO2; and 9 of
them CO. Data collected for PM10 and SO2 are daily averages measured in 3mgµ ; for
CO, the data is measured in ppm and is the maximum of 8 hours; and finally NO2 is
measured in 3mgµ and is the maximum of 1 hour.
It was calculated monthly averages for all pollutants on each monitoring station
and them for the whole city.
13
4.2 Hospital Visits Data
Hospital visits monthly data was provided by DATASUS for January 1998 until
March 2006. DATASUS is the SUS Hospital Information System, administered by the
Ministry of Health. At this system, the hospital units who belongs to SUS (both public
and private ones), provide information about the number hospital visits by nature of the
disease occurred in each month of the year. This information is grouped in accordance
with the International Diseases Classification.
It was collected data for children bellow 1 year old and children between 1 and 4
year old. It’s important to say that the SUS (Integrated Health System) is only
responsible for 50% of all the hospitals attendances in Brazil and most of the population
covered by it belong to the lowest income classes (Freitas, s/d)
4.3 Meteorological Data
Monthly minimum temperature and relative humidity data were provided by the
Astronomy, Geophysics and Atmospheric Sciences Institute (IAG/USP). The data
obtained from IAG/USP was considered representative of the city as a whole.
5. Results
This topic will present the variables analyzed as well as the results obtained from
the modeling. As already said before, the association between child visits and pollutants
concentration will be done through an additive generalized model.
To create a model, the variables listed bellow was used:
Monthly Hospital Visits due to Respiratory Diseases
• Children under 1 year old;
• Children between 1 and 4 years old;
•
14
Monthly data for Average Pollutants Concentration
• Particulate Matter of size less than 10 3mµg ;
• Carbon Monoxide;
• Nitrogen Dioxide;
• Sulphur Dioxide;
Monthly data for Meteorological Factors
• Minimum Temperature
• Humidity.
Figure 3 show the hospitals visits evolution from January 1998 until March
2006. It can be noticed the existence of a cycle in the series, because on winter’s month
the number of hospitals visits goes up meanwhile during the summertime it goes down.
0
2000
4000
6000
8000
10000
12000
jan/9
8
jul/9
8
jan/9
9
jul/9
9
jan/0
0
jul/0
0
jan/0
1
jul/0
1
jan/0
2
jul/0
2
jan/0
3
jul/0
3
jan/0
4
jul/0
4
jan/0
5
jul/0
5
jan/0
6
RSMP
Cidade de SP
Figure 2 – Total Hospital Visits due to Respiratory Diseases in São Paulo and
Metropolitan Area (number of people)
Source: DATASUS (2006)
The age groups were divided as children, teenagers, adults and elderly. It was
considered children people with less than 9 years old; teenagers people from 10 to 19
years old, adults people from 20 to 59 years old and elderly people with more than 60
years old. This classification has the intent of separating the active economic population
and point out that children are the ones with more susceptibility to respiratory diseases.
15
In 2005, 53% of all hospitals visit due to respiratory diseases were done by children and
elderly were responsible for 20,4% of it. In other words, the most susceptible groups
had been the cause of more than 70% of its hospitals visits.
0
500
1000
1500
2000
2500
3000
3500
4000
4500
jan/
98
jan/
99
jan/
00
jan/
01
jan/
02
jan/
03
jan/
04
jan/
05
jan/
06
Criancas
Jovens
Adultos
Idosos
Figure 3– Hospital Visits due to Respiratory Diseases by Age Groups in São Paulo
(number of people)
Source: DATASUS (2006)
Of all the respiratory disease hospital visits, children between 1 and 4 years old
had 21,6% of its responsibility, and the children with less than 1 year old 21%. In
Figure 3 and Figure 4 can be visualized the seasonal cycle, especially for children
(Figure 3).
0
500
1000
1500
2000
2500
jan/
98
jul/9
8
jan/
99
jul/9
9
jan/
00
jul/0
0
jan/
01
jul/0
1
jan/
02
jul/0
2
jan/
03
jul/0
3
jan/
04
jul/0
4
jan/
05
jul/0
5
jan/
06
menor de 1 ano
1 a 4 anos
mais de 60 anos
Figure 5 – Hospital Visits due to Respiratory Diseases in Children and Elderly People –
São Paulo
Source: DATASUS (2006)
16
From Figure 6 to Figure 9, are presented the pollutants concentration evolution.
At São Paulo, SO2 concentration decreased 46%, NO2 reduced 11%, CO had a 65%
reduction and PM10 a 32% decrease. Especially for PM10 and CO is possible to capture,
just by a glance in the figures, the seasonal component of the two series. The number of
studies in the literature has already mentioned the existence of this event in São Paulo
city (CETESB, 2006; Miraglia, 2002). All the pollutants concentrations are bellow the
annual averages limits determined by CONAMA Resolution no. 03 from 28/06/90
(CETESB, 2006).
25,00
35,00
45,00
55,00
65,00
75,00
85,00
jan/
98
jul/9
8
jan/
99
jul/9
9
jan/
00
jul/0
0
jan/
01
jul/0
1
jan/
02
jul/0
2
jan/
03
jul/0
3
jan/
04
jul/0
4
jan/
05
jul/0
5
jan/
06
Figure 6 – Particulate Matter (PM10) Concentration
Source: CETESB (2006)
0,00
5,00
10,00
15,00
20,00
25,00
30,00
jan/
98
jul/9
8
jan/
99
jul/9
9
jan/
00
jul/0
0
jan/
01
jul/0
1
jan/
02
jul/0
2
jan/
03
jul/0
3
jan/
04
jul/0
4
jan/
05
jul/0
5
jan/
06
Figure 7 – Sulphur Dioxide (SO2) Concentration
Source: CETESB (2006)
17
40,00
60,00
80,00
100,00
120,00
140,00
160,00
jan/
98
jul/9
8
jan/
99
jul/9
9
jan/
00
jul/0
0
jan/
01
jul/0
1
jan/
02
jul/0
2
jan/
03
jul/0
3
jan/
04
jul/0
4
jan/
05
jul/0
5
jan/
06
Figure 8 – Nitrogen Dioxide (NO2) Concentration
Source: CETESB (2006)
1,00
1,50
2,00
2,50
3,00
3,50
4,00
4,50
5,00
5,50
jan/
98
jul/9
8
jan/
99
jul/9
9
jan/
00
jul/0
0
jan/
01
jul/0
1
jan/
02
jul/0
2
jan/
03
jul/0
3
jan/
04
jul/0
4
jan/
05
jul/0
5
jan/
06
Figure 9 – Carbon Monoxide (CO) Concentration
Source: CETESB (2006)
At Table 3, it is possible to observe the humidity and minimum temperature. The
seasonal cycle is much clearer in the minimum temperature variable than on the
humidity. But a cycle exists in both of them.
Table 3 – Meteorological Data
Meteorological Variables 1998 1999 2000 2001 2002 2003 2004 2005 2006Humudity 81,70 79,71 79,46 79,47 78,83 74,04 76,86 77,07 78,39
Temperature 12,06 10,71 9,68 10,63 11,63 14,43 13,84 14,85 18,43
18
5.1 Descriptive Analysis
A descriptive analysis was made for the dependent and independent variables
(available on Table 4). The minimum temperature varied from 2oC to 18 oC,
approximately, and humidity from 73 to 85%. The child hospital visits monthly average
was 849 for children with less than 1 year old and 818 for children between 1 and 4
years old.
Table 4 – Descriptive Analysis
Average Mediana ModaStandard Deviation
Variance Minimum Maximum
Children under 1 year old 849 761 397 394 155.311 305 2173
Children between 1 and 4 year old
818 781 712 219 47.887 402 1365
Humidity 80,05 80,51 73,12 2,78 7,75 73,12 85,07Temperature 10,98 11,30 5,20 3,91 15,32 2,00 17,90PM10 45,83 43,00 35,27 11,81 139,46 28,00 75,91NO2 98,70 97,71 93,00 18,67 348,73 51,80 137,17SO2 13,46 13,25 12,00 3,93 15,42 6,66 24,17CO 3,05 2,97 2,53 0,89 0,80 1,32 5,13
Variables
Table 5 – Independent Variables Quartiles
25 50 75PM10 36,23 43,00 53,86NO2 86,67 97,71 112,68
SO2 10,6 13,25 16,68CO 2,4 2,97 3,61IOP 0,8313 0,9775 1,1334
Variaveis Quartis
Nitrogen dioxide was the one that had the highest variance meanwhile CO
concentration was the most stable one. The two maximum concentrations PM10 and
NO2 were above the annually limits.
Quartiles were calculated for the pollutants variables to be used at the dose-
dependence analysis and are presented on Table 5. The next step on the study is
establishing the correlation between the child hospital visits and its dependent variables.
19
A Spearman correlation was used to capture the association between child
hospital visits because of respiratory diseases and the independent variables (pollutant
concentration and meteorological factors). Table 6 shows the statistics obtained
Children under 1 year old presented strong positive association with PM10, NO2
and S02. No significant association was observed with CO. For children with age
between 1 and 4 years old, the significant association was only obtained with NO2.
Meteorological factors a strong negative association was found. This means that the
relation between child respiratory disease hospital visits and meteorological factors is
inverse. For example, when temperature has gone down, child hospital visits due to
respiratory diseases will go up.
Table 6 – Spearman Correlations
Variables PM10 SO2 NO2 COHumidity Minimum
Temperature (oC)
Children under 1 year old 0,40253 * 0,40592 * 0,4614 * 0,11521 -0,24823 -0,66346 *Children between 1 and 4 years old 0,16826 0,14763 0,36250 * -0,17872 -0,1862 -0,45884 ** p < 0,0083
Since pollutants concentration have multicolinearity problems, it was also
important to analyze the association between themselves. It could be noticed that all
pollutants concentrations have strong correlations. Particulate matter was the pollutant
which presented the strongest one. Table 7 shows the correlation between the pollutants.
Table 7 – Pollutants Correlation
Variables SO2 NO2 COPM10 0,724** 0,696** 0,788**
SO2 0,497** 0,602**
NO2 0,471**
5.2 Additive Generalized Models - AGM
The additive generalized model was the framework used to relate the children
hospital visits with pollutants concentration. At the first, it was created a model
containing all the pollutants, meteorological factors, and annual and monthly dummies.
20
Pollutants concentrations and meteorological data was smoothing through cubic splines
before being added to the model. The results are presented at Table 8 and Table 9.
Table 8 – Children under 1 year old Additive Generalized Models – Initial Model
Explicative Variables CoefficientStandard
Error z-value p-valueIntercept 6,0391 0,1927 31,3463 0Month
january -0,1848 0,0206 -8,9874 0february -0,3174 0,0215 -14,7669 0march -0,1326 0,021 -6,3007 0april 0,3103 0,0215 14,4419 0may 0,7638 0,0247 30,9549 0june 0,6946 0,0346 20,0624 0july 0,4722 0,034 13,8845 0
august 0,2322 0,0327 7,1122 0september 0,0877 0,0253 3,4625 0,00054
october 0,0162 0,0226 0,7147 0,47478november 0,0143 0,0198 0,7223 0,47012december
Years1998 -0,4671 0,0442 -10,5793 01999 -0,4472 0,0405 -11,0486 02000 -0,2847 0,0411 -6,92 02001 -0,2224 0,036 -6,1835 02002 -0,0579 0,0332 -1,7468 0,080682003 0,1719 0,0303 5,6836 02004 0,1564 0,0333 4,6985 02005 0,0659 0,0293 2,2502 0,024442006
Meteorological VariablesHumidity 0,0049 0,0021 2,3401 0,01928Temperature 0,0029 0,002 1,4623 0,14366PollutantsPM10 * 0,0009 0,0009 1,0211 0,30721SO2 * 0,0115 0,0022 5,2236 0
NO2 * 0,0007 0,0004 1,9078 0,05642CO * -0,0299 0,0147 -2,0324 0,04211
At this model SO2 was the pollutant which presented the highest association and
CO presented a protection characteristics. Using a backforward selection process, a
“final model” was obtained to explain the relation between the variables. Table 10 and
11 show the results for the final models.
21
Table 9 - Children Between 1 year and 4 years old Additive Generalized Models – Initial Model
Explicative Variables Coefficient Standard Error z-value p-valueIntercept 7,0819 0,1955 36,2308 0
Monthjanuary -0,2392 0,0193 -12,3846 0february -0,3185 0,02 -15,8841 0march -0,0455 0,0195 -2,3295 0,01983april 0,2114 0,021 10,0882 0may 0,2827 0,0253 11,1787 0june 0,3104 0,0355 8,7426 0july 0,2268 0,0348 6,5169 0
august 0,1261 0,0325 3,8814 0,0001september 0,0659 0,0246 2,6792 0,00738
october 0,0501 0,0215 2,3287 0,01987november 0,0219 0,0185 1,1843 0,23631december
Years1998 -0,2852 0,0433 -6,5912 01999 -0,3349 0,0391 -8,5555 02000 -0,1569 0,0395 -3,9721 0,000072001 -0,1649 0,0346 -4,7722 02002 -0,0573 0,0317 -1,8055 0,0712003 0,1188 0,0286 4,1562 0,000032004 0,0454 0,0318 1,4279 0,153332005 0,0844 0,0276 3,0602 0,002212006
Meteorological VariablesHumidity -0,0034 0,0021 -1,5696 0,11652Temperature 0 0,002 0,0083 0,99336PollutantsPM10 * -0,0014 0,001 -1,5105 0,13092SO2 * 0,002 0,0022 0,8781 0,37991NO2 * 0,0012 0,0004 3,3821 0,00072CO * -0,0583 0,0153 -3,808 0,00014 The AGM obtained for children under 1 year old shows that the number of child
hospitals visits increases at the end of autumn and begging of winter, in other words, in
May, June and July. At the initial model 4,21%, 16,75% and 7,15% of child respiratory
diseases hospital visits occurred due to PM10, SO2 and NO2 concentrations, respectively.
At the final model, on average 17% of child respiratory diseases hospital visits
happened because of SO2 emissions.
22
Table 10 - Children under 1 year old Additive Generalized Models – Final Model
Explicative Variables Coefficient Standard Error z-value p-valueIntercept 6,1692 0,1373 44,9232 0
Monthjanuary -0,2108 0,0202 -10,4583 0february -0,3611 0,0208 -17,3524 0march -0,1401 0,0201 -6,9746 0april 0,3297 0,0194 16,9687 0may 0,7676 0,0186 41,3376 0june 0,6648 0,02 33,2342 0july 0,4263 0,0208 20,5051 0
august 0,1986 0,0235 8,4406 0september 0,1077 0,0197 5,4558 0
october 0,0594 0,0196 3,0336 0,00242november 0,0568 0,0193 2,9466 0,00321december
Years1998 -0,4785 0,0303 -15,8181 01999 -0,474 0,0311 -15,2201 02000 -0,3168 0,0317 -9,9796 02001 -0,2438 0,0291 -8,3905 02002 -0,0642 0,0287 -2,2388 0,025172003 0,1796 0,0276 6,4951 02004 0,1522 0,0295 5,1552 02005 0,0699 0,0271 2,5793 0,00992006
Meteorological VariablesHumidity 0,0041 0,0016 2,4787 0,01318PollutantsSO2 * 0,0117 0,0021 5,6821 0 To children 1 and 4 years old, SO2 and NO2 were responsible for 2,75% and
11,59% of the respiratory diseases hospital visits. At the final model NO2 was
responsible for 9,29% respiratory diseases hospital visits. Despite that some statistics
obtained at the initial model was not considered statistically significant, it’s important to
take into consideration, so the next section will point out some numbers calculated from
the models.
5.3 Inter-Quartiles Analysis
Based on Table 5 (independent variables quartiles), it was calculated the
percentage of influence that pollutants have on child hospital visits because of
respiratory diseases. In fact, it was calculated the relative risk of each one of the
pollutant quartiles for the two age groups studied. Table 12 present that information.
23
Table 11 - Children between 1 and 4 years old Additive Generalized Models – Final Model
Explicative Variables Coefficient Standard Error z-value p-valueIntercept 6,7592 0,0379 178,455 0
Monthjanuary -0,2542 0,019 -13,377 0february -0,3449 0,0198 -17,456 0march -0,0605 0,0185 -3,273 0,00107april 0,2105 0,0185 11,364 0may 0,2849 0,0197 14,449 0june 0,2795 0,0271 10,3 0july 0,1846 0,0251 7,36 0
august 0,1053 0,0237 4,447 0,00001september 0,075 0,0194 3,872 0,00011
october 0,0743 0,0186 3,986 0,00007november 0,0464 0,0177 2,63 0,00853december
Years1998 -0,279 0,0382 -7,295 01999 -0,3093 0,0333 -9,284 02000 -0,1357 0,0327 -4,144 0,000032001 -0,1522 0,031 -4,908 02002 -0,042 0,0295 -1,424 0,154462003 0,1541 0,0269 5,736 02004 0,0824 0,0272 3,032 0,002432005 0,1213 0,0262 4,635 02006
PollutantsNO2 * 0,0009 0,0003 2,632 0,00848CO * -0,0465 0,012 -3,874 0,00011
At the initial model 35 thousand children visited the hospitals with respiratory diseases
caused by pollutants concentrations. At the final model 27 thousand child hospital visits
occurred.
24
Table 12 – Child Hospital Visits due to Respiratory Diseases Caused by Air Pollutants
Concentration – Initial and Final Models
25 50 75Inital Model - Respiratory Diseases Hospital Visits
Children under 1 year old PM10 3.316 2.785 3.316 4.174 NO2 5.948 5.256 5.948 6.896 SO2 13.831 10.893 13.831 17.768
Children between 1 and 4 years old NO2 10.078 8.880 10.078 11.730
SO2 2.176 1.736 2.176 2.748 Total 35.349 29.550 35.349 43.315 Final Model - Respiratory Diseases Hospital Visits
Children under 1 year oldSO2 14.125 12.227 13.812 15.966
Children between 1 and 4 years old NO2 5.355 4.661 5.239 6.016 Total 19.480 16.887 19.051 21.982
VariablesAverage
ConcentrationConcentration Quartiles
Comparing the results obtained with the ones available in the literature, it’s
possible to realize that the present results are in accordance with what has already been
concluded by other scientists. Table 13 shows the results of other studies.
Table 13 – Comparative Studies about the Effects of Air Pollution on Children
Respiratory Diseases Hospital Visits
Place Pollutant Studied Relative Risk
São Paulo (Braga et al, 1999) PM10 1,12
São Paulo (Lin et al, 1999) PM10 1,05
São Paulo (Farhat, 1999) PM10 1,08
São Paulo (Gouveia et al,2003) PM10 1,07CO 1,02SO2 1,07
Daejon (Cho et al, 2000) NO2 1,47
São Paulo Children under 1 year old SO2 1,01Children between 1 and 4 years old NO2 1,00
The study aimed to visualize the existing association between pollutants
concentration and child hospital visits. Despite of not always being statistically
significant important correlations were found with particulate matter, nitrogen dioxide
and sulphur dioxide. In future studies maybe the use of an index created as a fusion of
25
all the four pollutants concentration would be an important upgrade to epidemiologic
studies that deal with pollutants effect on human health.
Bibliography
Akerman, M., Campanário, P., Maia, P.B. Saúde e Meio Ambiente: análise de
diferenciais intra-urbanos. Município de São Paulo, Brasil. Revista de Saúde
Pública, v.30, (4), pp. 372-82, 1996.
Akimoto, Hajime. Global Air Quality and Pollution. Science, Dec 2003, vol. 302 www.sciencemag.org.
André, P. A. et al. Environmental Epidemiology Applied to Urban Atmospheric
Pollution: a contribution from the Experimental Air Pollution Laboratory (LPAE).
Caderno de Saúde Pública, v.16, (3), pp. 619-628, jul-set 2000.
Azevedo, L.A. P., Berenstein, M., Junior, F. F. C. Analysis of the Influence of Climate
Change and Atmospheric Pollutants on Respiratory Diseases in Children in an
Emergency Setting. Alergia, Asma e Inmunología Pedriátricas, v.8, (3), pp. 74-
84, maio-junho 1999.
Azuanga, Denise. Danos Ambientais Causados por Veículos Leves no Brasil. Rio de
Janeiro: COPPE, Universidade Federal do Rio de Janeiro, 2000, 126p. Tese
(Doutorado).
Bakonyi, S. M. C. et al. Poluição Atmosférica e Doenças Respiratórias em Crianças na
Cidade de Curitiba, PR. Revista de Saúde Pública, v.38, (5), pp. 695-700, 2004.
Barbosa, S. R. C. S. Desenvolvimento e Ambiente: Questões Fundamentais da
Sociologia Contemporânea. Humanitas, v.3, (2), pp. 39-53, 2000.
Barquera, S., et al. Metodologia na pesquisa epidemiológica de doenças respiratórias e
poluição ambiental. Revista de Saúde Pública, v.36, (1), pp. 107-113, 2002.
Botter, D. A., Jørgensen, B., Peres, A.A.Q. A Longitudinal study of mortality and air
pollution for São Paulo, Brazil. Journal of Exposure Analysis and Environmental
Epidemiology, v.12, pp. 335-343, 2002.
Braga, A., Pereira, L. A. A., Saldiva, P. H. N. Poluição Atmosférica e Seus Efeitos na
Saúde Humana.
CETESB, Dados Fornecidos pelo DETRAN/PRODESP. São Paulo: 2005.
26
CETESB. Relatório de Qualidade do Ar no Estado de São Paulo. São Paulo: Gráfica
CETESB, 2006. 90p. Disponível em:
www.cetesb.gov.br/Ar/relatorios/relatorios.asp
Cho, B., Choi, J., Yum, Y. Air Pollution and Hospital Admissions for Respiratory
Disease in Certain Areas of Korea. Journal of Occupational Health, v.42, pp.
185-191, 2000.
Clean Air Initiative. To understand how the health impacts of air pollution are
measured and valued. 39p.
Conceicão, G.M.S. et al. Air Pollution and Child Mortality: a time series study in São
Paulo, Brazil. Environmental Health Perspectives, v.109,(supl 3), pp.347-350,
2001.
Conceição, G.M.S., Saldiva, P. H. N., Singer, J. M. Modelos MLG e MAG para análise
da associação entre poluição atmosférica e marcadores de morbi-mortalidade: uma
introdução baseada em dados da cidade de São Paulo. Revista Brasileira de
Epidemiologia, v.4,(3), pp.206-219, 2001.
Daumas, Regina Paiva. Poluição do Ar e Mortalidade em Idosos no Município do Rio
de Janeiro: Análise de Serie Temporal. Rio de Janeiro: Instituto de Medicinal
Social, Universidade do Estado do Rio de Janeiro, 2002, 104p. Tese (Mestrado).
DETRAN/PRODESP (Departamento de Análises). Arquivo: Frota Circulante – 2003,
São Paulo, 2004.
Fleiss, J. L. Statistical Methods for rate and Proportions. Ed: John Wiley Sons, Inc,
Second Edition, 1981.
Freitas, C. U., Pereira, L.A.A., Saldiva, P.H.N. Vigilância dos Efeitos na Saúde
Decorrentes da Poluição Atmosférica: Estudo de Factibilidade.
Francisco, P. M. S. B. et al. Tendência de Mortalidade por Doenças Respiratórias em
Idosos do Estado de São Paulo, 1980 a 1998. Revista de Saúde Pública, v.37, (2),
pp. 191-6, 2003.
Gouveia, N. et al. Poluição do Ar e Saúde em Duas Grandes Metrópoles Brasileiras na
Década de 90. Informe Epidemiológico do SUS, v.11, (1), pp. 41-43, 2002.
Grant, G.R. Ewens, W. J. Statistical Methods in Bioinformatics. New York: Springer.
2001. 475p.
Hartman, R. S., Wheeler, D., Singh, M. The Cost of Air Pollution Abatement.
Dezembro,1994. 45p.
27
Hogan, D., Berquó, E., Costa, H. S. M., Population and Environment in Brazil: Rio
+10. Campinas: CNPD, ABEP, NEPO, 2002, Health Effects of Ambient Level of
Air Pollution, pp. 207-224.
Hogan, D. et al. Migração e Meio Ambiente em São Paulo: Aspectos Relevantes da
Dinâmica Recente. Campinas: Editora da Unicamp, Um Breve Perfil Ambiental
do Estado de São Paulo, pp. 275-381.
IBAMA. Programa de controle da poluição do ar por veículos automotores –
PROCONVE. Brasília, 2ed, 1998.
Jacobi, P.R, Macedo, L.V. Consciência dos cidadãos e poluição atmosférica na região
metropolitana de São Paulo – RMSP. CEPAL Serie Medio Ambiente y Desarollo.
2001. 44p.
Kleinman, A. Et al. Social Suffering. California, The Appeal of Experience; The
Dismay of Images: Cultural Appropriations of Suffering in our times, pp. 1-23.
Martins, L.C. et al Poluição Atmosférica e Atendimentos por Pneumonia e Gripe em
São Paulo, Brasil. Revista de Saúde Pública, v.36, (1), pp. 88-94, 2002.
Medeiros, M. A., Augusto, A. Menos mortes pelo ar. Pesquisa FAPESP, São Paulo, pp. 36-39, out. 2001.
Miraglia, Simone Georges El Khouri. Análise do Impacto do Consumo de Diferentes
Combustíveis na Incidência de Mortalidade por Doenças Respiratórias no
Município de São Paulo. São Paulo: Escola Politécnica, Universidade de São
Paulo, 1997, 100p. Tese (Mestrado).
Miraglia, Simone Georges El Khouri. O Ônus da Poluição Atmosférica sobre a
População do Município de São Paulo. São Paulo: Faculdade de Medicina,
Universidade de São Paulo, 2002, 126p. Tese (Doutorado).
Miraglia, S.G.E.K., Saldiva, P. H. N., Böhm, G. M. An Evaluation of Air Pollution
Health Impacts and Costs in São Paulo, Brazil. . Environmental Management,
v.35, (5), pp. 667-676, 2005.
Montibeller-Filho,G. O Mito do Desenvolvimento Sustentável. Meio Ambiente e Custos
Locais no Moderno Sistema Produtor de Mercadorias. Florianópolis: Ed. Da
UFSC, 2001. 306p.
Morin, E. A Cabeça Bem-feita. Repensar a reforma, reformar o pensamento. Rio de Janeiro: Bertrand Brasil, 2000: 9 – 33.
Motta, R.S., Mendes, A. P. F. Custos de Saúde Associados à Poluição do Ar no Brasil.
Texto Para Discussão no. 332. Brasília: IPEA, 1994. 32p.
28
Motta, R. S., Ferreira, S. F., Ortiz, R. A. Valoração Econômica dos Impactos Causados
pela Poluição Atmosférica na Saúde Humana: Estudos para a Região
Metropolitana de São Paulo. Projeto CETESB/ Banco Mundial/ USP/ IPEA. Rio
de Janeiro: 1998. 9p.
Nascimento, L. F. C et al. Efeitos da Poluição Atmosférica na Saúde Ambiental Infantil
em São José dos Campos. Revista de Saúde Pública, v.40, (1), pp. 77-82, 2006.
Pagano, M., Gauvreau, K. Princípios de Bioestatística. São Paulo: Pioneira Thomson
Learning, segunda edição, 2004.
Panyacosit, L. A Review of Particulate Matter and Health: Focus on Developing
Countries. 38p.
Paula, G. A. Modelos de Regressão com Apoio Computacional. Instituto de Matemática
e Estatística. Universidade de São Paulo. 245p.
O’Neil, M. S. et al. Health, Wealth and Air Pollution: Theory and Methods.
Environmental Health Perspectives, v.111, (16), pp. 1861-1870, 2003.
Ostro, B. D. et al Air Pollution and Health Effects: a Study of Respiratory Illness
Among Children in Santiago, Chile. Marco 1998.
Pagano, M. Gauvreau, K. Princípios de Bioestatística. São Paulo: Pioneira Thomson
Learning. 2004.
Riediker, M. et al. Particulatte Matter Expsure in Car is Associated with Cardiovascular
Effects in Health Young Men. American Journal of Respiratory and Critical Care
Medicine, v.169, pp. 934-940 2004.
Rio, E M. B., Gallo, P. R., Siqueira, A. A. F. Mortalidade por Asma no Município de
São Paulo, Brasil. Revista de Saúde Pública, v.36, (2), pp. 149-54, 2002.
Romieu, I. et al. Impactos de la Contaminación Atmosférica em la Mortalidad y
Morbilidad de la Población Infantil de Ciudad Juaréz, Chihuahua. Mexico.
Montreal: 2003. 38p.
Romieu, I., Borja-Burto, V.H. Particulate Air Pollution and Daily Mortality: Can
Results be Generalizad to Latin American Countries? Salud Publica de México,
v.39, (5), pp. 1-9, 1997.
Rumel, D. et al. Infarto do Miocárdio e Acidente Vascular Cerebral Associados à Alta
de Temperatura e Monóxido de Carbono em Área Metropolitana do Sudeste do
Brasil. Revista de Saúde Pública, v.27, (1), pp. 15-22, 1993.
29
Samet, J. M. et al. Research Report: The National Morbidity, Mortality and Air
Pollution Study. Part II: Morbidity and Mortality from Air Pollution in the US.
Health Effects Institute. v.94,(2), pp. 1-82.
Silva, E P., et al. Recursos Energéticos, Meio Ambiente e Desenvolvimento. 22p.
Simas, Hugo Segrilo. Aspectos Metodológicos em Análise de Series Temporais
Epidemiolólicas do Efeito da Poluição Atmosférica na Saúde Pública: uma
revisão bibliográfica e um estudo comparativo via simulação. Rio de Janeiro:
Instituto de Medicinal Social, Universidade do Estado do Rio de Janeiro, 2003,
107p. Tese (Mestrado).
Stieb, D. M., Judek, S., Burnett, R. T. Meta-analysis of time-series studies of air
pollution and mortality: effects of gases and particles and the influence of cause of
death, age and season. Journal of the Air & Waste Management Association, v.52,
pp.470-484, 2002.
Switzer, P., Lefohn, A. S. A Review of the Statistical Methodologies Applied in the
Time-Series Epidemiologic Studies of Ambient Carbon Monoxide and Health.
April, 1999.
Trunk, M. T., Figueiras, A., Lareo, I. C. Efectos a Corto Plazo de la Contaminación
Atmosférica sobre la Mortalidad. Resultados del Proyecto EMECAM en la
Ciudad de Vigo, 1996-04. Revista Española de Salud Publica, v.73, pp.275-282,
1999.
Warren, J. E., Grant, G.R. Statistical Methods in Bioinformatics. Estados Unidos:
Springer, 2001. 476p.
WHO. Air Quality Guidelines for Particulate Matter, Ozone, Nitrogen Dioxide and
Súlfur Dioxide:Global Update 2005. Geneva: WHO Press, 2005. 20p.
Woolson, R.F. Statistical Methods for the Analysis of Biomedical Data. Estados Unidos:
John Willey & Sons, 1987. 513p.
World Bank Group. The Effects of Pollution on Health: The Economic Toll. Pollution
Prevention and Abatment Handbook. July, 1999, pp.63-71.
Zhang, J. et al. Children’s Respiratory Morbidity Prevalence in Relation to Air
Pollution in Four Chinese Citites. Environmental Health Perspectives, v.110,(9),
2002.